Outliers

What are outliers in Omnia, and how are they defined?

Sometimes a price shown in Omnia can deviate from the rest. This can have multiple causes: the price on the shop is incorrect, or the price is incorrectly scraped.

Either way, we do not want to use these prices when calculating our calculated fields. 

In the product & pricing information tab, an outlier is shown by a red exclamation mark: 

How does Omnia calculate outliers?

Omnia calculates whether an article is an outlier based on all received price points that are available for a retailer. This means we might also take into account prices that are not directly shown in your product overview, but that are in our database because we scrape these price points for other customers. Doing this improves our accuracy when calculating outliers because it lets us assess price points on all the information we have available beyond internal customer data. However, and for statistical accuracy purposes, Omnia needs at least four price points to be able to calculate an outlier. 

Two checks are done to calculate our outliers:

Price point compared to the mean
First we look at a price point relative to the mean of all available price points for an article within a country. If the price point is relatively far away from the mean, we say that the deviation is too big. We therefore flag this price as an outlier. 

Absolute check of price difference
Second, we check whether the absolute difference between a price point and the mean is also big. This reduces the chance that we incorrectly flag prices with a small deviation as an outlier. 

What is the impact of an outlier?

If a price is flagged as an outlier, we do not take into account that price point when calculating our calculated fields. So average prices, minimum prices, etc., skip a price point if that price is flagged as an outlier.